short report interlaboratory test on the method un test n ...test+o… · test sample, approx. 250...

Short Report Interlaboratory test on the

method UN Test N.5 / EC A.12 “Substances which, in contact with

water, emit flammable gases” 2007

Berlin, 28.11.2011

Dipl.-Math. Kirstin Kunath

Dr. Peter Lüth PD Dr. habil. Steffen Uhlig

reported by

QuoData GmbH – Quality Management and Statistics BAM Federal Institute for Materials Research and Testing

Authors

Dipl.-Math. Kirstin Kunath 1 Dr. Peter Lüth 2 PD Dr. habil. Steffen Uhlig 1 1 QuoData GmbH 2 BAM Federal Institute for Materials Research and Testing

Impressum Interlaboratory test on the method UN Test N.5 / EC A.12 “Substances which, in contact with water, emit flammable gases” 2007 Short Report Herausgeber: BAM Bundesanstalt für Materialforschung und -prüfung Unter den Eichen 87 12205 Berlin Telefon: +49 30 8104-0 Telefax: +49 30 8112029 Internet: www.bam.de Copyright © 2011 by BAM Bundesanstalt für Materialforschung und -prüfung ISBN 978-3-9814634-1-5

Organisation Panel

Center for quality assurance for testing of dangerous goods and hazardous substances

Operation & Administration

BAM Federal Institute for Materials Research and Testing

Dr. Peter Lüth

Unter den Eichen 87

D-12205 Berlin

Phone: +49 (0)30-81041201

Fax: +49 (0)30-81041207

Email: [email protected]

Statistical Design, Analysis and Evaluation

QuoData GmbH PD Dr. habil. Steffen Uhlig, Dipl.-Math. Kirstin Kunath

Kaitzer Straße 135

D-01187 Dresden

Phone: +49 (0)351-4028867-0

Fax: +49 (0)351-4028867-19

Email: [email protected]

Short Report

BAM Federal Institute for Materials Research and Testing

Dr. Peter Lüth

Unter den Eichen 87

D-12205 Berlin

QuoData GmbH

PD Dr. habil. Steffen Uhlig, Dipl.-Math. Kirstin Kunath

Kaitzer Straße 135

D-01187 Dresden

Contents 1 Introduction ........................................................................................................................................ 1

2 Aim ..................................................................................................................................................... 1

3 Test material ...................................................................................................................................... 2

3.1 Interlaboratory test sample “metal mixture powder” .................................................................... 2

3.1.1 Preparation of the interlaboratory test samples ................................................................. 2

3.1.2 Homogeneity and stability of the interlaboratory test samples .......................................... 4

4 Procedure of the interlaboratory test.................................................................................................. 4

4.1 Organisation ................................................................................................................................ 4

4.2 Test procedure ............................................................................................................................ 5

4.2.1 Interlaboratory test instruction and laboratory data input form .......................................... 6

5 Evaluation .......................................................................................................................................... 7

5.1 Background ................................................................................................................................. 7

5.2 Quantity of test results ................................................................................................................. 7

5.3 Plausibility check ......................................................................................................................... 7

5.4 Results and resulting classification of the interlaboratory test sample “Metal mixture powder” ....................................................................................................................................... 7

5.5 Kernel density estimation of single pressure rise times ............................................................10

5.6 Performance of the method UN Test N.5 / EC A.12 .................................................................11

5.6.1 Reproducibility, repeatability and classification error ......................................................11

5.6.1.1 Probability of incorrect classification for an arbitrary sample ............................. 11

5.6.2 Zu scores ..........................................................................................................................13

5.8 Other influencing (disturbing) factors ........................................................................................15

6 Summary and conclusions ...............................................................................................................17

6.1 Performance of the methods UN Test N.5 / EC A.12 ................................................................17

6.2 Influencing factors .....................................................................................................................17

6.3 Recommendations for the participants of the interlaboratory test to improve execution of the method ................................................................................................................................18

6.4 Recommendations to improve execution of the method ...........................................................19

7 References .......................................................................................................................................20

8 Appendix ..........................................................................................................................................21

8.1 Test of the homogeneity and stability ........................................................................................21

8.2 Test instruction ..........................................................................................................................28

8.3 Laboratory data input form ........................................................................................................29

List of tables Table ‎4-1: Participating laboratories (n=12) .......................................................................................... 4

Table ‎4-2: Time schedule of the interlaboratory test ............................................................................. 5

Table ‎4-3: Test methods of the Round Robin test ................................................................................ 5

Table ‎4-4: Scope of the methods .......................................................................................................... 6

Table ‎5-1: Results of the Laboratory – single values and laboratory mean value ................................ 8

Table ‎5-2: Results of the Laboratory – Robust mean value and standard deviation over all

laboratories (according to DIN 38402-45 ‎[7]) ....................................................................... 9

Table ‎5-3: Classification results of the interlaboratory test sample “Metal mixture powder”

according to the methods UN Test N.5 / EC A.12 based on the maximum value of

the gas evolution rate (i.e. maximum of the three single values) (robust mean value

and robust reproducibility s.d. according to DIN 38402-45 ‎[7]) ........................................... 9

Table ‎5-4: Zu scores according to DIN 38402-45 / ISO/IEC 17043 of the laboratories ......................13

Table ‎6-1: Recommendations to improve the execution of the method .............................................18

List of figures Figure ‎3-1: Dividing of the 16.6 kg stock sample on a spinning (rotating) riffler at BAM right

picture: the sample flows from the hopper to a vibrating chute and travels along the

chute to the receivers (8 glass containers) .......................................................................... 2

Figure ‎3-2: Division of the sample of the cross riffling procedure .......................................................... 3

Figure ‎5-1: Gas evolution rates (singe values) of all laboratories .......................................................... 8

Figure ‎5-2: Kernel Density Estimation of the gas evolution rates (Reference value: 3.18 +/-

0.71) with: bars denote the tolerance limits (|Z|=2) (as an example for laboratory

910) triangle denotes the laboratory mean (as examples for laboratory 910) ..................10

Figure ‎5-3: Analysis of gas evolutions rates of all laboratories with tolerance limits (red) and

with: SR…reproducibility s.d., Sr….repeatability s.d. .......................................................11

Figure ‎5-4: Shark profile – probability of incorrect classification in PG III or no classification as a

function of the "true" gas evolution rate of an arbitrary sample .........................................12

Figure ‎5-5: Overview of the zu scores of the laboratories ....................................................................14

Figure ‎5-6: Gas evolution rate of laboratories with different measuring devices: G Gravimetry

V Volumetry V-MS Volumetry with magnetic stirrer V-PG Volumetry with a

pressure gauge V-GFM Volumetry with gas flow meter V-AB Volumetry with

automated gas burette .......................................................................................................15

Figure ‎5-7: Ambient temperature during the detection of the gas evolution rate in different

laboratories ........................................................................................................................16

Figure ‎5-8: Relationship between the gas evolution rate and the ambient temperature of

laboratories using “Volumetry (V)” (6 of 12 laboratories) ...................................................16

Interlaboratory test 2007 on the method UN N.5 / EC A.12 – Introduction

1

QuoData GmbH / BAM

1 Introduction

For the classification of chemicals, special standardized test procedures have been developed and are

used world-wide. Safe handling and use of these chemicals depend on the correct classification which

therefore must be based on the precise and correct execution of the tests and their evaluation. In this

context interlaboratory tests (round robin tests, interlaboratory comparisons / intercomparisons) are a

crucial element of a laboratory's quality system. Participation in interlaboratory tests is explicitly rec-

ommended by the standard ISO/IEC 17025.

The present document reports on the results of the interlaboratory test 2007 on the method UN Test

N.5 “Test method for substances which in contact with water emit flammable gases” [1] / EC A.12

“Flammability (contact with water)” [2] which was organized by the Center for Quality Assurance for

Testing of Dangerous Goods and Hazardous Substances.

In dependence on the chemical structure and/or the physical form and state (e.g. particle size) sub-

stances or mixtures may be able to react with water (even water damp / air humidity) under normal

ambient temperature conditions. Sometimes this reaction can be violent and/or with significant genera-

tion of heat. Especially if gases are evolved this reaction may become dangerous. In addition, it is

important to know whether a substance emits flammable gases due to contact with water because

special precautions are necessary especially with regard to explosion protection.

The methods UN Test N.5 and EC A.12 are applied to characterize chemical substances or mixtures

which in contact with water emit flammable gases. To differentiate between chemicals with these

properties and chemicals which are not classified as hazardous / dangerous, the substance’s gas

evolution rate is determined and compared to the classification criteria(s) in the last step of the test

method. In the methods UN Test N.5 / EC A.12 no special laboratory apparatus / measuring technique

to determine gas evolving flow is required.

However, practical experience shows that the testing procedure for substances and mixtures which in

contact with water emit flammable gases is sensitive to a number of influencing factors.

Since the methods (UN N.5 / EC A.12) were developed and came into force in the early nineties a

systematic review concerning the practical application of the test method has not been carried out.

2 Aim

The aims of this interlaboratory test 2007 on the method UN Test N.5 / EC A.12 “Substances which, in

contact with water, emit flammable gases” are:

1. Assessment of the performance of the methods UN Test N.5 / EC A.12: The current practical application of the method in different laboratories and the classification error

were assessed by the interlaboratory test. For this purpose specific precision indicators (e.g. re-

producibility, repeatability, probability of incorrect classification etc.) were generated.

2. Assessment of other influencing factors: Other laboratory specific factors which possibly may have an influence on the test re-

sult / classification were evaluated with the aid of a further exploratory analysis.

Interlaboratory test 2007 on the method UN N.5 / EC A.12– Test material

2

QuoData GmbH / BAM

3. Recommendations for the participants of the interlaboratory test to improve execution of the method:

In view of the results of the interlaboratory test it was assessed which recommendations could be

given to the participating laboratories to improve execution of the method.

3 Test material

3.1 Interlaboratory test sample “metal mixture powder”

“Metal mixture powder” (Zn and ZnO (< 87%), Al and Al2O3 (appr.9%), Mg and MgO (appr.3%)) was

chosen as the interlaboratory test sample.

3.1.1 Preparation of the interlaboratory test samples

In February 2007 the original stock sample (16,6 kg) was divided into 64 sub-samples (interlaboratory

test sample, approx. 250 g) by aid of a spinning (rotating) riffler (see Figure 3-1) [3].

Figure 3-1: Dividing of the 16.6 kg stock sample on a spinning (rotating) riffler at BAM right picture: the sample flows from the hopper to a vibrating chute and travels along the chute to the receivers (8 glass containers)

The homogenising procedure was performed in accordance with the principles of the cross-riffling-

procedure [4] (see Figure 3-2).

Interlaboratory test 2007 on the method UN N.5 / EC A.12 – Test material

3

QuoData GmbH / BAM

Initial volume

1. step

1 2 3 4 5 6 7 8 8 sub-samples (1-8)

2. step

Cross-riffling

1-1 2-2 3-3 4-4 5-5 6-6 7-7 8-8 A

1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-1 B

1-3 2-4 3-5 4-6 5-7 6-8 7-1 8-2 C

1-4 2-5 3-6 4-7 5-8 6-1 7-2 8-3 D

1-5 2-6 3-7 4-8 5-1 6-2 7-3 8-4 E

1-6 2-7 3-8 4-1 5-2 6-3 7-4 8-5 F

1-7 2-8 3-1 4-2 5-3 6-4 7-5 8-6 G

1-8 2-1 3-2 4-3 5-4 6-5 7-6 8-7 H

A B C D E F G H 8 sub-samples (A-H)

3. step

A-1 B-2 C-3 D-4 E-5 F-6 G-7 H-8 Recombination

A-2 B-3 C-4 D-5 E-6 F-7 G-8 H-1

A-3 B-4 C-5 D-6 E-7 F-8 G-1 H-2

A-4 B-5 C-6 D-7 E-8 F-1 G-2 H-3

A-5 B-6 C-7 D-8 E-1 F-2 G-3 H-4

A-6 B-7 C-8 D-1 E-2 F-3 G-4 H-5

A-7 B-8 C-1 D-2 E-3 F-4 G-5 H-6

A-8 B-1 C-2 D-3 E-4 F-5 G-6 H-7

Figure 3-2: Division of the sample of the cross riffling procedure

The divided interlaboratory test samples were packed into transport containers and sent to the partici-

pants of the interlaboratory test.

Interlaboratory test 2007 on the method UN N.5 / EC A.12– Procedure of the interlaboratory test

4

QuoData GmbH / BAM

3.1.2 Homogeneity and stability of the interlaboratory test samples

Based on experience with mixed metal powders and the interlaboratory test sample “Metal mixture

powder” and the test methods the stock sample are known to be sufficiently homogeneous and stable

within the testing time frame of the round robin test (June 2007 to September 2007).

In addition, tests of the homogeneity and stability were performed. For this purpose the gas evolution

rate of several interlaboratory test samples was checked in the period from February 2007 until Octo-

ber 2007 (e.g. before sending out the test samples, in the middle of the testing period and after the

end of the testing period. The test results are listed in Annex 8.1. The expected homogeneity and sta-

bility of the interlaboratory test samples were confirmed by the results of these additional tests.

4 Procedure of the interlaboratory test

4.1 Organisation

The interlaboratory test was organized by BAM Federal Institute for Materials Research and Testing,

Berlin, in the frame of the interlaboratory test programme within the Center for Quality Assurance for

Testing of Dangerous Goods and Hazardous Substances.

The interlaboratory test sample “Metal mixture powder” together with a test instruction and the labora-

tory data input form (see Appendix 8.2 and 8.3), was distributed to 12 participating laboratories (see

Table 4-1).

Table 4-1: Participating laboratories (n=12)

Laboratory / Agency Country

AIST National Institute of Advanced Industrial Science and Technology Japan

ALMAMET GmbH Germany

AQura GmbH Germany

BAM Federal Institute for Materials Research and Testing Germany

BASF AG Germany

Bayer Industry Services GmbH & Co. OHG Germany

Chilworth Technology Ltd. United Kingdom

Government Laboratory of Hong Kong HKSAR China

Henkel AG & Co. KGaA Germany

INERIS France

Minton, Treharne & Davies Ltd. United Kingdom

Siemens AG Germany

Interlaboratory test 2007 on the method UN N.5 / EC A.12 – Procedure of the interlaboratory test

5

QuoData GmbH / BAM

The following Table 4-2 shows the time schedule of the study.

Table 4-2: Time schedule of the interlaboratory test

Round Robin step Time period

Public announcement May 2007

Order and registration June 2007

Distribution of the test sample and the test instruction June 2007

Laboratory testing period July 2007 to September 2007 *

First statistical evaluation and draft report November to December 2007 **

Distribution of the Certificates December 2007

* ...... The testing period was prolonged from August 2007 to September 2007 because not all labora-tories were able to perform the tests until August 2007 (as originally foreseen in June 2007).

** ..... The statistical evaluation was postponed due to the prolonged testing period.

4.2 Test procedure

The interlaboratory test was performed on the methods UN Test N.5 “Test method for substances

which in contact with water emit flammable gases” and EC A.12 “Flammability (contact with water)”

(see Table 4-3):

Table 4-3: Test methods of the Round Robin test

Test method Source

UN Test N.5 “Test method for substances which in contact with water emit flammable gases”

United Nations: Recommendations on the Transport of Dangerous Goods, Manual of Tests and Criteria, Fifth revised edition, United Nations, New York and Geneva, 2009 [1]1

EC A.12 “Flammability (contact with water)”

European legislation: Council Regulation (EC) No 440/2008 of 30 May 2008 [2]

Both methods, UN Test N.5 and EC A.12, are comparable with regard to the general principles and

the criteria for classification and non-classification.

In contrast to EC A.12, UN Test N.5 does not only distinguish between substances and mixtures

which are classified as flammable and those that are not but it also divides classified substances and

mixtures into three categories / packing groups (see Table 4-4).

1 UN Test N.5 has not been changed compared to the Fourth revised edition of the UN Manual of Tests and Criteria which

was effective at the time of the interlaboratory test.

Interlaboratory test 2007 on the method UN N.5 / EC A.12– Procedure of the interlaboratory test

6

QuoData GmbH / BAM

Table 4-4: Scope of the methods

Method Purpose

UN Test N.5 “Test method for substances which in contact with water emit flammable gases”

Decision on classification in accordance with - GHS: Classification in hazard class

"Substances and mixtures which, in con-tact with water, emit flammable gases" and assignment of the appropriate cate-gory 3, 2 or 1

- TDG: Classification in Division 4.3 and assignment of the appropriate Packing group III, II or I

EC A.12 “Flammability (contact with water)” Decision on classification as F; R15

4.2.1 Interlaboratory test instruction and laboratory data input form

The test instruction for the interlaboratory test (see Appendix 8.2) was distributed together with the test

sample.

The test instruction is more detailed than the current description of UN Test N.5 and EC A.12 and

included the following information:

The interlaboratory test sample should be tested as received, i.e. according to the following in-

structions:

- DO NOT treat the test sample by any mechanical method before testing

(no sieving, no grinding).

- In case that the test sample is slightly agglomerated, crush possible agglomerates only very

gently (by hand).

Preparation of test samples with a precision of 10.0 g ± 0.1 g.

Using of distilled (or demineralized) water.

The amount of distilled (or demineralized) water per test shall be 20.0 ml ± 0.5 ml.

Performing of the procedure as prescribed in chapter 33.4.1.4.3.5 of UN Test Manual (4th rev. edi-

tion, 2003) or, respectively, in chapter 1.6.4 of the European method A.12.

Performing the test over a period of 8 hours after adding the water, i.e. the rate of the evolu-

tion of gas should be calculated over 7 hours at 1 hour intervals.

The test should be performed in triplicate.

Apart from that, the other details of the procedure were supposed to be applied as usual in the labora-

tory and in accordance with chapter 34.4.1.4.3.5 of the UN Manual of Tests and Criteria [1] or/and in

chapter 1.6.4 of the European method EC A.12 [2].

Laboratory specific parameters and test conditions were enquired with the aid of the laboratory data

input form (see Appendix 8.3).

Interlaboratory test 2007 on the method UN N.5 / EC A.12 – Evaluation

7

QuoData GmbH / BAM

5 Evaluation

5.1 Background

The evaluation of the data was performed using a specially modified version2 of the software package

ProLab Plus 2007 [5]. ProLab Plus is widely employed for the evaluation of interlaboratory tests and

laboratory proficiency tests.

The method according to DIN 38402-45 [7] (=ISO/TS 20612) was applied for the statistical analysis of

gas evolution rates (see remark below Figure 5-2). This method is a robust method and no outlier

examination is required.

5.2 Quantity of test results

Measurements were conducted by 12 laboratories and all of them submitted results (see Table 4-1).

5.3 Plausibility check

To begin with, a first check of the plausibility of the data was performed. All results of the trials were

considered as plausible regarding the gas evolution rate (e.g. with regard to completeness, consisten-

cy etc.).

5.4 Results and resulting classification of the interlaboratory test sample “Metal mixture powder”

The laboratory single values and the mean value of the gas evolution rate for the interlaboratory test sample are given in the following Table 5-1 and Table 5-2.

2 The basic ProLab Plus version has been extended by additional tools taking into account the specific design of the intercom-parison. These additional tools are in-house tools only.

Interlaboratory test 2007 on the method UN N.5 / EC A.12– Evaluation

8

QuoData GmbH / BAM

Table 5-1: Results of the Laboratory – single values and laboratory mean value

Laboratory Gas evolution rate [l / (kg x h)]

Value 1 Value 2 Value 3 Laboratory mean value

066 3.9 3.6 3.5 3.67

106 1.9 1.6 1.9 1.80

121 1.53 1.53 1.63 1.56

201 0.4 2.1 0.5 1.00

251 3.1 3.6 3.4 3.37

272 3.38 2.85 3.67 3.20

902 4.2 4.2 4.2 4.20

906 3.6 3.3 4.0 3.63

907 5.7 5.6 4.3 5.20

908 2.22 2.95 2.2 2.46

910 3.97 4.13 3.97 4.02

911 3.5 4.3 3.8 3.87

Note: The number of decimals was not prescribed by the interlaboratory test instruction and therefore are differ-ent between laboratories.

The laboratory single values of the gas evolution rate of the Interlaboratory test sample are shown in

Figure 5-1.

Figure 5-1: Gas evolution rates (singe values) of all laboratories

The robust mean value and standard deviation of the gas evolution rates over all laboratories is given

in Table 5-2.


9

QuoData GmbH / BAM

Table 5-2: Results of the Laboratory – Robust mean value and standard deviation over all laboratories (according to DIN 38402-45 [7])

Parameter Value [gas evolution rate [l / (kg x h)]]

Robust mean value over all laboratories 3.18

Robust reproducibility s.d. over all laboratories 1.24

Based on these test results, the centrally distributed interlaboratory test sample “Metal mixture pow-

der” would be classified by all laboratories in the same hazard class "Substances and mixtures which,

in contact with water, emit flammable gases", category 3 and in the same division 4.3, packing group

(PG III) (see Table 5-3).

Table 5-3: Classification results of the interlaboratory test sample “Metal mixture powder” according to the methods UN Test N.5 / EC A.12 based on the maximum value of the gas evolution rate (i.e. maximum of the three single values) (robust mean value and robust reproducibility s.d. according to DIN 38402-45 [7])

Laboratory / Kind of value

Maximum gas evolution rate [ l /(kg x h)]

Classification result

066 3.9 Division 4.3, PG III / F; R15












Robust mean value of the max-imum gas evolution rates 3.45 l / (kg x h)

Robust reproducibility standard deviation of the maxi-mum gas evolution rates 1.04 l / (kg x h)

Interlaboratory test 2007 on the method UN N.5 / EC A.12– Evaluation

10

QuoData GmbH / BAM

5.5 Kernel density estimation of single pressure rise times

An analysis of the underlying distribution of the single gas evolution rates was carried out by the so-

called kernel density estimation in order to check the homogeneity. The following figure Figure 5-2

shows the results of the kernel density estimation for all single gas evolution rates.

The blue curve characterizes the distribution of single gas evolution rates obtained by the kernel den-

sity estimation.

The y-axis of the kernel density plot shows the probability density. This probability density is neither

the probability nor the frequency. It indicates the relative frequency of values occurring at different

points along the x-axis. Thereby not the y-axis values are of interest, but only the shape of the curve.

Figure 5-2: Kernel Density Estimation of the gas evolution rates

(Reference value: 3.18 +/- 0.71) with: bars denote the tolerance limits (|Z|=2) (as an example for laboratory 910) triangle denotes the laboratory mean (as examples for laboratory 910)

Remark: According to the Kernel Density Plot the data are not normally distributed. Therefore

DIN ISO 5725-2 cannot be applied. Even though the method according to DIN ISO 5725-5 is more

robust than the method according to ISO 5725-2, the method described in DIN ISO 5725-5 is more

sensitive to within-lab outliers than the method according to DIN 38402-45 [7]. Furthermore, in case

the interlaboratory test is repeated, the method according DIN ISO 5725-5 is less efficient in terms of

the reliability of the precision parameters (mean, standard deviation) than the method according to

DIN 38402-45. Therefore DIN 38402-45 is used in this evaluation.

Gas evolution rate [l/(kg x h)]876543210-1-2

Den

sity

1.6

1.4

1.2

1

0.8

0.6

0.4

0.2

0

0.69 5.663.18

3.844.21

4.02


11

QuoData GmbH / BAM

5.6 Performance of the method UN Test N.5 / EC A.12

5.6.1 Reproducibility, repeatability and classification error

A summary of the obtained mean values, repeatability and reproducibility standard deviations of the

laboratory gas evolutions rates is shown in the following Figure 5-3.

In this figure, the size of the blue boxes symbolizes the laboratory's relative repeatability standard

deviation. The horizontal line in the middle of the box indicates the laboratory's mean gas evolution

rate. The figures also include the robust mean of the single gas evolution rates according to

DIN 38402-45 (black line) as well as the limits of tolerance (red) – also according to DIN 38402-45.

Figure 5-3: .. Analysis of gas evolutions rates of all laboratories with tolerance limits (red) and with:

SR…reproducibility s.d., Sr….repeatability s.d.

The mean gas evolution rate of laboratory 201 is very low compared to other laboratories and below

the lower tolerance.

5.6.1.1 Probability of incorrect classification for an arbitrary sample False positive error of classification: A false positive error indicates that the test sample is classi-

fied by a laboratory to a lower packing group (of higher safety), although a higher packing group (of

lower safety) would be correct because this is the “true” packing group.

False negative error of classification: A false negative error indicates that the test sample is classi-

fied in a higher packing group (of lower safety), although a lower packing group (of higher safety)

would be necessary.

The probability of incorrect classification can also be predicted for an arbitrary test sample mixture, i.e.

not only for the test sample which was used in the interlaboratory test. [3]

Figure 5-4 shows the probability of incorrect classification for an arbitrary test sample as tested under

Laboratory

201

121

106

908

272

251

906

066

911

910

902

907

Gas

evo

lutio

n ra

te [l

/(kg

x h)

]

7

6

5

4

3

2

1

0

SR

Sr


12

QuoData GmbH / BAM

the conditions of this interlaboratory test. Due to the form of the curve, such figures are also referred to

as shark profiles. They can be interpreted as follows:

The increasing arm of a curve of the probability of incorrect classification with respect to a certain

packing group indicates the probability of the false positive classification (false positive error) as a

function of the true gas evolution rate, i.e. the probability that the test sample is incorrectly classified in

a lower packing group (of higher safety) than it correctly should be, based on its "true" pressure rise

time (x-value).

On the other hand, the decreasing arm of a curve of one packing group indicates the probability of the

false negative classification (false negative error), i.e. the probability that the test sample is incorrectly

classified in a higher packing group (of lower safety) than it correctly should be, based on its "true"

pressure rise time (x-value).

In particular, this means:

Left arm of packing group III (increasing arm)

Probability that a laboratory classifies a test sample in PG III, although no classification would be correct (true) false positive classification

Right arm of packing group III (decreasing arm)

Probability that a laboratory does not classify a test sample, although PG III would be necessary (true) false negative classification

Read the Shark Profile as follows:

If for example the "true" gas evolution rate of an arbitrary test sample mixture amounts to

2.5 l / (kg x h), the probability for a false negative classification with regard to packing group III, i.e. the

probability to not classify the test sample instead of the “true” packing group III, is about 9 %.

Figure 5-4: Shark profile – probability of incorrect classification in PG III or no classification as a func-

tion of the "true" gas evolution rate of an arbitrary sample

False positive classification False negative classification


13

QuoData GmbH / BAM

5.6.2 Zu scores

The assessment of the laboratory’s performance in determining the gas evolution rate of the laborato-

ry test sample was carried out using zu scores according to DIN 38402-45 / ISO/IEC 17043. In gen-

eral, zu scores describe the standardised deviation of laboratory mean values from the total mean

value over all laboratories (= robust mean value over all laboratories, see Table 5-2) under considera-

tion that the lower limit of tolerance does not fall below zero. Under normal distribution, zu scores lie

within the limits -2 and 2 with probability 95 % and therefore if |zu score| > 2 holds, the quality criterion

is not fulfilled.

Summarised, a laboratory’s result is

satisfactory if ........... |zu score| ≤ 2;

questionable if ....... 2 < |zu score| < 3;

unsatisfactory if ....... |zu score| ≥ 3.

In general, a zu score smaller (greater) than zero means that the laboratory’s mean is smaller (greater)

than the total mean over all laboratories (= robust mean value over all laboratories, see Table 5-2).

In this interlaboratory test one of the zu scores (laboratory 201) lies outside the limits of tolerance (see

Table 5-4 and Figure 5-5).

Table 5-4: Zu scores according to DIN 38402-45 / ISO/IEC 17043 of the laboratories

Laboratory zu score

066 0.31

106 -1.31

121 -1.53

201 -2.07

251 0.12

272 0.01

902 0.65

906 0.29

907 1.28

908 -0.68

910 0.54

911 0.44


14

QuoData GmbH / BAM

zu score

Figure 5-5: Overview of the zu scores of the laboratories

-2 0 2

P1/GASENT

Labo

rato

ry066

106

121

201

251

272

902

906

907

908

910

911

-2,065


15

QuoData GmbH / BAM

5.8 Other influencing (disturbing) factors

An additional exploratory analysis of the gas evolution rates was conducted in order to find the reason

for the broad spread of the results and the deviations between the laboratories.

Different measuring devices und techniques are used in the laboratories (see Figure 5-6).

Figure 5-6: Gas evolution rate of laboratories with different measuring devices: G Gravimetry V Volumetry V-MS Volumetry with magnetic stirrer V-PG Volumetry with a pressure gauge V-GFM Volumetry with gas flow meter V-AB Volumetry with automated gas burette

Two laboratories have used a gravimetric detection technique. It is remarkable that the results of these

two laboratories are very similar and have a comparatively high repeatability.

Very low gas evolution rates were determined by laboratory 201 which measures the gas evolution

rate by volumetry and uses a magnetic stirrer in addition. It therefore might be concluded that the use

of a stirrer considerably lowers the measured gas evolution rate.

Laboratory

201

121

106

908

272

251

906

066

911

910

902

907

Gas

evo

lutio

n ra

te [l

/(kg

x h)

]

6

5

4

3

2

1

0

V-MS

V-PG

V-G

FM V

V V V V

V G G

V-AB


16

QuoData GmbH / BAM

Furthermore, the ambient room temperature in the laboratories was different (see Figure 5-7).

Figure 5-7: Ambient temperature during the detection of the gas evolution rate in different laboratories

The relationship between the gas evolution rate and the ambient temperature is shown in Figure 5-8

for those 6 laboratories using "Volumetry (V)".

Figure 5-8: Relationship between the gas evolution rate and the ambient temperature of laboratories

using “Volumetry (V)” (6 of 12 laboratories)

No significant relationship can be identified between the ambient temperature and the gas evolution

rate for laboratories with comparable test techniques (Volumetry).

Interlaboratory test 2007 on the method UN N.5 / EC A.12 – Summary and conclusions

17

QuoData GmbH / BAM

6 Summary and conclusions

The following conclusions can be drawn concerning the aim of this interlaboratory test.

The methods do not prescribe a certain procedure or apparatus for determining the gas evolution rate.

As a result the laboratories use different procedures and equipment where the main difference is ap-

plication of a volumetric or a gravimetric method. As demonstrated by this interlaboratory test the gas

evolution rates measured by the different laboratories vary in a wide range.

Additionally, the test methods do not foresee the use of a reference material.

Therefore it is recommended that adequate quality control measures are consistently applied.

6.1 Performance of the methods UN Test N.5 / EC A.12

Even though the interlaboratory test sample “Metal mixture powder” would be classified by all labora-

tories in the same hazard class "Substances and mixtures which, in contact with water, emit flamma-

ble gases", category 3 and in the same division 4.3, packing group (PG III) (see Table 5-3), the gas

evolution rates of the single values varied in a wide range from 0.4 to 5.7 l / (kg x h). However, it must

be noted that the interlaboratory test on the method UN Test N.5 / EC A.12 shows significant differ-

ences in the determination of the gas evolution rates and a relatively high probability of incorrect clas-

sification (see chapter 5.6.1.1).

Further, it was observed that the participants of the interlaboratory test 2007 used different devices to

measure the gas evolution rate like classical volumetric measuring devices (e.g. gas burette or auto-

matic gas burette) as well as other measuring techniques / devices (e.g. pressure gauge or gas flow

meter or gravimetric detection technique). It is remarkable that the results of the two laboratories

which used a gravimetric detection technique are very similar and have a comparatively high repeata-

bility. On the other hand, very low gas evolution rates were determined by the laboratory which

measures the gas evolution rate by volumetry and uses a magnetic stirrer in addition. It therefore

might be concluded that the use of a stirrer considerably lowers the measured gas evolution rate.

It is recommended that the reasons for these differences in the measured gas evolution rates when

applying test method UN Test N.5 / EC A.12 are investigated in further and more specific investiga-

tions.

6.2 Influencing factors

An additional exploratory analysis of the gas evolution rates was conducted in order to find the reason

for the broad spread of the results and the deviations between the laboratories.

Based on the results of one laboratory it might be concluded that the use of a stirrer considerably low-

ers the gas evolution rate.

It was found that ambient temperature has no influence of the gas evolution rate.


18

QuoData GmbH / BAM

6.3 Recommendations for the participants of the interlaboratory test to improve execution of the method

In view of the results of the interlaboratory test the following recommendations for improving execution

of the method can be given to the participating laboratories (see Table 6-1).

Table 6-1: Recommendations to improve the execution of the method

Laboratory Application recommendation

066 No additional recommendations. 106 Even though the result of laboratory 106 was "satisfactory" according to DIN

38402-45 the following recommendations are given: 1. Check the accuracy of the measuring device “Gas flow meter FM4 (SYSTAG

SystemTechnik AG, Switzerland)” by comparison with an independent method (e.g. glass burette).

2. Check compliance with the method with regard to the requirement that the test is to be performed at 20 °C.

121 Even though the result of laboratory 121 was "satisfactory" according to DIN 38402-45 the following recommendations are given: 1. Check the accuracy of the measuring device “pressure gauge” by comparison

with an independent method (e.g. glass burette). 2. Check compliance with the method with regard to the requirement that the test

is to be performed at 20 °C. 201 1. Check the influence of the procedure stirring with “magnetic stirrer” by compari-

son with a test performed without stirring the sample. Remark: The method gives no recommendation to use a stirrer.

2. Check compliance with the method with regard to the requirement that the test is to be performed at 20 °C.

251 Even though the result of laboratory 251 was "satisfactory" according to DIN 38402-45 the following recommendation is given: 1. Check compliance with the method with regard to the requirement that the test

is to be performed at 20 °C. 272 Even though the result of laboratory 272 was "satisfactory" according to DIN

38402-45 the following recommendation is given: 1. Check compliance with the method with regard to the requirement that the test

is to be performed at 20 °C. 902 No additional recommendations. 906 Even though the result of laboratory 906 was "satisfactory" according to DIN


is to be performed at 20 °C. 907 Even though the result of laboratory 907 was "satisfactory" according to DIN

38402-45 the following recommendations are given: 1. Check the accuracy and the precision of the measuring device “automated gas

burette system” by comparison with an independent method (e.g. glass burette). 2. Check compliance with the method with regard to the requirement that the test

is to be performed at 20 °C. 908 Even though the result of laboratory 908 was "satisfactory" according to DIN 38402-

45 the following recommendation is given: 1. Check compliance with the method with regard to the requirement that the test

is to be performed at 20 °C. 910 No additional recommendations. 911 Even though the result of laboratory 911 was "satisfactory" according to DIN


is to be performed at 20 °C.


19

QuoData GmbH / BAM

It is recommended that quality control measures, which were applied to check the accuracy of the

laboratory results, should be noted in the laboratory test reports.

6.4 Recommendations to improve execution of the method

Based on the interlaboratory test, the experience gained and the actual results, the following

measures / actions are recommended:

1. Training of personnel:

Special samples should be manufactured and distributed centrally that may be used by

the laboratories for their internal quality control (e.g. RM (reference material) or CRM (cer-

tified reference material)).

An appropriate proficiency test scheme should be developed in order to be used for regu-

lar external quality control.

2. Development of the method UN Test N.5 / EC A.12

Identification of the causes for the differences between the laboratories with the aid of

more specific investigations (cause study).

It should be considered to agree on a specific procedure and equipment for determination

of the gas evolution rate and the methods UN Test N.5 / EC A.12 should be revised ac-

cordingly and give a clear and unmistakable description of the method.

Interlaboratory test 2007 on the method UN N.5 / EC A.12 – References

20

QuoData GmbH / BAM

7 References

[1] Recommendations on the Transport of Dangerous Goods, Manual of Tests and Criteria, Fifth

revised edition, United Nations, New York and Geneva, 2009.

[2] Council Regulation (EC) No 440/2008 of 30 May 2008 laying down test methods pursuant to

Regulation (EC) No 1907/2006 of the European Parliament and of the Council on the Registra-

tion, Evaluation, Authorisation and Restriction of Chemicals (REACH), OJ L 142, 31.5.2008.

[3] Antoni S, Kunath K, Lüth P, Schlage R, Simon K, Uhlig, S, Wildner W, Zimmermann C (2011)

Evaluation of the interlaboratory test on the method UN test O.1 "Test for oxidizing solids" with

sodium perborate monohydrate 2005/06. (PDF, 2011) BAM Berlin, ISBN 978-3-9814281-2-4

[4] van der Veen AMH, Nater DAG (1993) Sample preparation from bulk samples – an overview;

Fuel Processing Technology 36, 1-7

[5] ProLab Plus 2007 – The software package for method interlaboratory tests and proficiency

tests; specially modified version of the software package, QuoData GmbH, Dresden.

[6] Uhlig S, Lischer P (1998) Statistically-based performance characteristics in laboratory perfor-

mance studies; Analyst, 123, 167-172.

[7] DIN 38402-45: German standard methods for the examination of water, waste water and

sludge — General information (group A) — Part 45: Interlaboratory comparisons for proficiency

testing of laboratories (A 45), September 2003.

http://www.bam.de/de/service/publikationen/publikationen_medien/un_test_for_oxidizing_solids_final_report_on_interlab_test.pdf

http://www.bam.de/en/service/publikationen/onlinepublikationen.htm

Interlaboratory test 2007 on the method UN N.5 / EC A.12 – Appendix

21

QuoData GmbH / BAM

8 Appendix

8.1 Test of the homogeneity and stability


22

QuoData GmbH / BAM


23

QuoData GmbH / BAM


24

QuoData GmbH / BAM


25

QuoData GmbH / BAM


26

QuoData GmbH / BAM


27

QuoData GmbH / BAM


28

QuoData GmbH / BAM

8.2 Test instruction


29

QuoData GmbH / BAM

8.3 Laboratory data input form

short report interlaboratory test on the method un test n ...test+o… · test sample, approx. 250...

Documents